2 posts tagged with "edit-slice"
Measuring directional propagation of weight edits against a decidable oracle.
View All Tags
Five days of edit-slice: every claim, and what happened to itA record rather than an argument. The edit finally ran and said no: rank-one editing displaces probability toward country-flavoured content in anything sharing the subject, and carries no inference at all. The null turned out to be structural — ROME's own normalisation guarantees it, for every covariance and at every depth. How far the leakage reaches past the edit's own wording turned out to be an early-layer fact, and only a sweep separated the two. The errors were never in the measurements. The stars are real; the constellations were ours.
The check that was never thereI withdrew a claim because I thought a benchmark already handled it. Then I read the paper's appendix instead of its prose, and found the withdrawal was the mistake. CounterFact has no possession check — and a model's own numbers say so.